linux-kernel - Re: [PATCH 2/2] powerpc/topology: Override cpu_smt

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 04 Aug 2020 12:02:49 +0100
From:   Valentin Schneider <valentin.schneider@....com>
To:     peterz@...radead.org
Cc:     Srikar Dronamraju <srikar@...ux.vnet.ibm.com>,
        Ingo Molnar <mingo@...nel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Mel Gorman <mgorman@...hsingularity.net>,
        Rik van Riel <riel@...riel.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Vincent Guittot <vincent.guittot@...aro.org>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        linuxppc-dev <linuxppc-dev@...ts.ozlabs.org>,
        Michael Ellerman <mpe@...erman.id.au>,
        Michael Neuling <mikey@...ling.org>,
        Gautham R Shenoy <ego@...ux.vnet.ibm.com>,
        Vaidyanathan Srinivasan <svaidy@...ux.ibm.com>
Subject: Re: [PATCH 2/2] powerpc/topology: Override cpu_smt_mask


On 04/08/20 11:46, peterz@...radead.org wrote:
> On Tue, Aug 04, 2020 at 09:03:07AM +0530, Srikar Dronamraju wrote:
>> On Power9 a pair of cores can be presented by the firmware as a big-core
>> for backward compatibility reasons, with 4 threads per (small) core and 8
>> threads per big-core. cpu_smt_mask() should generally point to the cpu mask
>> of the (small)core.
>>
>> In order to maintain userspace backward compatibility (with Power8 chips in
>> case of Power9) in enterprise Linux systems, the topology_sibling_cpumask
>> has to be set to big-core. Hence override the default cpu_smt_mask() to be
>> powerpc specific allowing for better scheduling behaviour on Power.
>
> Why does Linux userspace care about this?

Ditto; from [1], a core contains CPUs that all share the same L1 (and capacity,
as per SD_SHARE_CPUCAPACITY). So IMO it makes perfect sense to have a first
domain spanning L1, and its parent spanning L2 - that means
topology_sibling_cpumask *itself* should span a single core rather than a
pair.

[1]: https://lkml.kernel.org/r/jhjr1sviswg.mognet@arm.com